October 18, 2001 9:00 AM PDT
Bell Labs makes molecule-size transistor
On Thursday, the research arm of Lucent Technologies unveiled a new organic transistor--made from carbon, rather than silicon--created by the chemical self-assembly of molecules. The transistor has a channel length of just a single molecule.
The prototype shows the potential, the researchers said, to allow for the creation of computer chips that are faster, more compact and easier to manufacture.
More research on the topic lies ahead, but experts agree that so-called nanotechnology will replace current, silicon-based processes sometime in the next decade, as the dimensions at which chips are built shrink and as silicon loses its ability to shrink with them. Despite silicon's limitations, though, manufacturing breakthroughs and new materials could always extend its life span.
A wide range of companies are investing for the post-silicon future.
IBM, for example, believes that the best bet for building non-silicon chips is the carbon nanotube, a tiny cylinder made of carbon. The computing stalwart's labs have already begun working with simple circuits based on those nanotubes.
Bell Labs has a long history in transistor research, including the invention of the device itself. Three of its scientists--William Shockley, John Bardeen and Walter Brattain--invented the transistor in 1947, an accomplishment that earned them the Nobel Prize for Physics nine years later.
Present Bell Labs researchers suggest that the organic transistor is a proof of principle, showing that transistors can be built on a molecular scale and also be used in future chips.
The lab has already assembled a prototype voltage inverter, an electronic circuit commonly used in chips to convert zeros to ones and vice versa--that is, to speak the basic language of computers.
Smaller means faster
The Bell Labs organic transistor features a channel length--the distance between electrodes--of one molecule, or about 1 to 2 nanometers, which is 100 times smaller than today's limit of 130 nanometers, according to the lab's researchers. Channel length determines the speed at which transistors can switch on and off. Additionally, because of their smaller size, thousands more transistors could be packed into a single chip, Jan Hendrick Schon, a researcher at Bell Labs, said in a conference call.
To put things into perspective, Intel's Pentium 4 chip is built using a 180-nanometer (0.18-micron) process, and it contains some 42 million transistors.
Constructing chips using this self-assembly process is also radically different than what the silicon world is used to now.
"We can do this process in a beaker, so we don't need a sophisticated clean room, and also to define this very short length scales, we don't need very sophisticated tools for lithography as used in nowadays' electronic circuits," Schon said.
The new transistors literally assemble themselves when a solution of organic semiconductor is poured on a base. The semiconductor molecules find and attach to the electrodes on their own, the researchers explained. Today's circuits are drawn on a silicon substrate, which is then etched away to form structures such as channels.
"It seems plausible that this is a much lower-cost and much easier process. But it's different. Anytime you introduce something that's kind of radically different, it requires a significant engineering development...to understand and build a device that has 10 million of these things connected up," said John Rogers, director of Bell Labs' Nanotechnology Research Department, in the same call.
Before the process could be used to build the kind of processors that power a PC, it would more likely be employed in flexible electronics or so-called plastic chips first.
"Low-cost plastic chips would be maybe the more immediate potential opportunity," Roberts said. "An exciting application we've been interested in here at Bell Labs for a couple of years now has been electronic paper-like displays."
These displays require flexible, durable drive circuitry, for which the new Bell Labs organic transistor is a good fit.
But much more research must be done, Bell Labs researchers stressed, before the organic transistor or its self-assembly techniques come close to production.